core/evals/deduction_serve/report.json
Shay b938ad617f feat(deduction-serve): Band v3-MEM — member-chain band, Socrates syllogism decided (ADR-0258)
Per-individual propositional lowering over minted (individual, class)
atoms: singular membership/predicate facts + all/every/each/no universals,
instantiated at every named individual, decided by the same verified ROBDD
engine. Closed-table number linking (irregular/invariant priority + three
regular suffix rules) is the one semantic identification — attested-pair
linking only, under-link over over-link. Composer tier strictly after
v1/v1b/v2-EN: monotone widening, v1 lane byte-identical.

- 4 new shape-bands (en_member_single/chain/negative/atomic) each earned
  SERVE at the arena: 13 bands x 720 = 9,360 wrong=0; ledger re-sealed.
- New hand-authored lane evals/deduction_serve/v2_member (26/26); ds-en-0022
  promoted declined->entailed (the designed acceptance case); report
  re-pinned; contract.md refreshed to the 4-band cascade.
- Typed refusals: quantifier pronouns, bare plurals, definite descriptions,
  relative clauses, tense, mixed connectives, universal conclusions.
- Flag posture unchanged (deduction_serving_enabled default off).

[Verification]: deductive 127 / smoke 180 / warmed_session 10; arena
all_bands_serve_licensed wrong_is_zero; lanes v1 28/28, v2_en 26/26,
v2_member 26/26.
2026-07-23 19:21:41 -07:00

82 lines
1.5 KiB
JSON

{
"aggregate": {
"correct": 80,
"declined": 0,
"n": 80,
"wrong": 0
},
"all_correct": true,
"arc": "deduction-serve",
"lane": "deduction_serve",
"schema_version": 1,
"splits": {
"v1": {
"all_cases_correct": true,
"by_gold": {
"declined": 4,
"entailed": 10,
"invalid": 1,
"refuted": 5,
"unknown": 5,
"valid": 3
},
"correct_by_gold": {
"declined": 4,
"entailed": 10,
"invalid": 1,
"refuted": 5,
"unknown": 5,
"valid": 3
},
"counts": {
"correct": 28,
"declined": 0,
"wrong": 0
},
"n": 28
},
"v2_en": {
"all_cases_correct": true,
"by_gold": {
"declined": 5,
"entailed": 14,
"refuted": 3,
"unknown": 4
},
"correct_by_gold": {
"declined": 5,
"entailed": 14,
"refuted": 3,
"unknown": 4
},
"counts": {
"correct": 26,
"declined": 0,
"wrong": 0
},
"n": 26
},
"v2_member": {
"all_cases_correct": true,
"by_gold": {
"declined": 7,
"entailed": 12,
"refuted": 3,
"unknown": 4
},
"correct_by_gold": {
"declined": 7,
"entailed": 12,
"refuted": 3,
"unknown": 4
},
"counts": {
"correct": 26,
"declined": 0,
"wrong": 0
},
"n": 26
}
},
"wrong_is_zero": true
}